Mustafa Can received his PhD from University College London (UCL), where he conducted research within the SpaceTimeLab on big data analytics and machine learning for intelligent transportation systems. His research focuses on advancing foundation models for spatio-temporal and complex network systems, with particular emphasis on graph-centric learning under network structural change and data sparsity.

He has developed integrated frameworks combining Graph Neural Networks, transformer-based models, and reinforcement learning to enable scalable and adaptive prediction in large-scale transportation networks. He also holds an MSc in Transport from Imperial College London and UCL, where he specialized in data-driven traffic modeling.

During his doctoral studies, he held teaching roles at UCL and the London School of Economics and Political Science (LSE), contributing to courses in data science, machine learning, and geospatial analysis. He also served as a Visiting Research Assistant at Yale University, working on foundation models for transport networks within the Computer Science Department.

His research lies at the intersection of spatio-temporal learning, graph representation learning, foundation models, and reinforcement learning, with applications in intelligent transportation systems and large-scale network systems.

• Spatio-temporal modeling and time series forecasting (traffic flow, multivariate data, transformers, foundation models)
• Graph Neural Networks and network science (dynamic graphs, structural changes, spatio-temporal GNNs)
• Foundation models for network systems (transformer and GNN-based approaches)
• Reinforcement learning and imitation learning (behavior cloning, decision-making, adaptive control)
• Data-driven network optimization and predictive modeling (scalable deep learning for large-scale systems)
• Transportation systems and geospatial data analysis (urban mobility, sensor data fusion, anomaly detection)

• Transformer and GNN-based foundation models for spatio-temporal data
• Integration of reinforcement learning with graph-based models
• Prediction and optimization in large-scale transportation networks
• Robust and anomaly-aware learning from real-world sensor data
• Scalable AI methods for evolving network systems